Table 4.

Multinomial logistic regression for univariate and multivariate association of changes of radiographic features from baseline to followup with mild and moderate pain trajectory (marginal pain trajectory is used as reference group). Values are OR (95% CI) unless otherwise specified.

VariablesMild Pain vs Marginal PainModerate Pain vs Marginal Pain
OR UnivariateOR Multivariate KL Model BOR Multivariate OP Model COR UnivariateOR Multivariate KL Model BOR Multivariate OP Model C
KL change
  Δ0ReferentReferentReferentReferentReferent
  Δ11.3 (0.9–2.0)1.4 (0.8–2.0)1.2 (0.8–1.9)1.1 (0.6–1.9)1.1 (0.6–1.9)
  Δ2 + Δ31.7 (0.9–3.5)*2.0 (0.9–4.5)*2.3 (1.1–4.7)**2.9 (1.2–7.1)**2.8 (1.0–7.4)**
Osteophyte area change1.0 (1.0–1.0)1.0 (1.0–1.1)1.0 (1.0–1.1) *1.0 (1.0–1.1)*
Min JSW change0.9 (0.8–1.1)0.9 (0.8–1.1)
Hosmer-Lemeshow goodness-of-fit test0.3220.3270.3220.327
Pseudo Nagelkerke R20.2260.2280.2260.228
  • * p value ≤ 0.1, cutoff point in univariate regression.

  • ** p value ≤ 0.05. KL change: radiographic change from baseline to followup in index knee. OP area change: OP change from baseline to 5-year followup in index knee. Min JSW change: change in minimal JSW from baseline to followup in index knee. Multivariate KL model B: model corrected for baseline variables (BMI, education, hip pain, worrying, resting, comorbidity, OP size, KL grade) and KL change. Multivariate OP model C: model corrected for baseline variables (BMI, education, hip pain, worrying, resting, comorbidity, OP size) and KL change, change of OP size. KL: Kellgren-Lawrence grading scale; OP: osteophyte; JSW: joint space width; BMI: body mass index.